Thermal Analyses of Sodalite, Tugtupite, Danalite and Helvite

163

The Canadian Mineralogist Vol. 40, pp. 163-172 (2002)

THERMAL ANALYSES OF SODALITE, TUGTUPITE, DANALITE AND HELVITE

SYTLE M. ANTAO Department of Geosciences, State University of New York, Stony Brook, New York 11794-2100, U.S.A.

ISHMAEL HASSAN¤ Department of Chemistry, University of the West Indies, Mona, Kingston 7, Jamaica

ABSTRACT

Differential thermal analyses (DTA) and thermogravimetric analyses (TG) were carried out on sodalite, tugtupite, danalite, and helvite using a Netzsch STA 409 EP/3/D simultaneous TGÐDTA apparatus. In the DTA and TG experimental run, a weighed amount of finely powdered sample was heated from 20 to 1450°C at a constant rate of 5°C/min. Sodalite, Na8[Al6Si6O24]Cl2, melts at 1079°C. The NaCl component in the sodalite-derived melt is lost in two stages; about 4.5 wt.% of NaCl is lost slowly at about 1150°C, and about 7.0 wt.% of NaCl is lost at a faster rate at about 1284°C. The interpretation that the NaCl component behaves this way in sodalite is confirmed by DTA and TG analyses of halite. Tugtupite, Na8[Al2Be2Si8O24]Cl2, melts at 1029°C. The NaCl component in tugtupite is lost in two main stages; about 1.8 wt.% of NaCl is first lost at about 1007°C, and about 8.2 wt.% of NaCl is lost in several steps between about 1018 and about 1442°C. Danalite, ideally Fe8[Be6Si6O24]S2, and helvite, 2+ 3+ ideally Mn8[Be6Si6O24]S2, undergo an oxidation of (Mn, Fe) to (Mn, Fe) cations. This is followed by a loss of S2(g), and then melting at about 1060°C. Finally, there is another oxidation of Mn3+ to Mn4+. These oxidations occur because the samples were heated in a static air environment. The first oxidation in danalite and helvite begins at about 771 and 705°C, respectively. Danalite gains 4.0 wt.%, whereas helvite gains 5.0 wt.% in the first weight-gain stage. In the weight-loss stage, danalite loses 5.7 wt.% of S2(g) from about 1029°C, whereas helvite loses 4.7 wt.% of S2(g) from about 883°C. The second stage of oxidation begins at about 1300°C in both the danalite- and helvite-derived melts. At this stage, the weight gain in the danalite-derived melt is about 0.7%, whereas that in the helvite-derived melt is about 0.2%.

Keywords: sodalite, tugtupite, danalite, helvite, differential thermal analysis, thermogravimetric analysis.

SOMMAIRE

Nous avons étudié la sodalite, la tugtupite, la danalite et la helvite par analyses thermique différentielle (ATD) et thermogravimétrique (TG) en utilisant un appareil Netzsch STA 409 EP/3/D. Pour ces expériences, nous nous sommes servis d’une quantité connue d’échantillon finement pulvérisé, chauffé à partir de 20 jusqu’à 1450°C à un taux constant de 5°C/min. La sodalite, Na8[Al6Si6O24]Cl2, atteint son point de fusion à 1079°C. La composante NaCl du bain fondu se volatilise en deux étapes; environ 4.5% (poids) de NaCl est perdu lentement près de 1150°C, et environ 7.0% est perdu plus rapidement près de 1284°C. Notre interpretation à propos de la sodalite est confirmée par analyses ADT et TG de la halite. La tugtupite, Na8[Al2Be2Si8O24]Cl2, fond à 1029°C. La composante NaCl de la tugtupite se volatilise en deux étapes; environ 1.8% de NaCl se perd près de 1007°C, et environ 8.2% de NaCl se perd en plusieurs étapes entre 1018 et 1442°C. La danalite, dont la composition idéale est 2+ 3+ Fe8[Be6Si6O24]S2, et la helvite, Mn8[Be6Si6O24]S2, subissent une oxydation de (Mn, Fe) à (Mn, Fe) . Cette transformation est 3+ 4+ suivie d’une perte de S2(g), et de la fusion à environ 1060°C. Enfin, il y a une oxydation plus poussée de Mn à Mn . Ces étapes d’oxydation sont le résultat du chauffage des échantillons dans l’air, dans un milieu statique. La première étape d’oxydation de la danalite et de la helvite débute à environ 771 et 705°C, respectivement. La danalite augmente en poids de 4.0%, tandis que la helvite augmente de 5.0%. Suit ensuite une perte de poids, 5.7% sous forme de S2(g) pour la danalite à partir de 1029°C, et 4.7% de S2(g) à partir de 883°C pour la helvite. Le second stade d’oxydation débute à environ 1300°C dans les liquides dérivés de la danalite et de la helvite. L’augmentation en poids à ce stade est d’environ 0.7% pour le cas de la danalite et 0.2% pour le cas de la helvite.

(Traduit par la Rédaction)

Mots-clés: sodalite, tugtupite, danalite, helvite, analyse thermique différentielle, analyse thermogravimétrique.

¤ E-mail address: [email protected]

163 40#1-fév-02-2328-11 163 3/28/02, 16:26 164 THE CANADIAN MINERALOGIST

INTRODUCTION measure the temperature where changes take place, and to compare the results for sodalite and tugtupite. If NaCl Our aim in this study is to determine the thermal is liberated, then it would be of interest to compare the behavior of sodalite, tugtupite, danalite, and helvite us- results for sodalite and tugtupite with those for halite, ing differential thermal analysis (DTA) and thermogra- NaCl. For this purpose, halite also was analyzed by the vimetry (TG) techniques. The types of volatiles liberated differential thermal analysis (DTA) and thermogravi- and the temperatures at which liberation occurs are de- metric analysis (TG) techniques. termined, among other features. From the chemical composition of danalite and In sodalite and tugtupite, Cl and possibly Na are helvite, S is an important volatile that could be liberated important volatile constituents that may be liberated on on heating. Therefore, this study on danalite and helvite heating. If the formula of sodalite is written as was carried out to determine the temperature where Na8[Al6Si6O24]Cl2, then Cl2 (g) could be expected to be S2(g) is liberated, and to determine the temperatures at liberated on heating; however, if the sodalite formula is which other structural changes may occur. written as Na6[Al6Si6O24]¥2NaCl, then NaCl could be liberated on heating; similarly for tugtupite, Na8[Al2Be2 REVIEW OF THE STRUCTURAL ASPECTS Si8O24]Cl2, which can be written as Na6[Al2Be2Si8 O24]¥2NaCl. This study was carried out to determine The frameworks of tugtupite, danalite, helvite, which chemical constituents are liberated on heating, to lazurite, nosean, and haüyne, which has the sodalite 4+ 3+ structure, consist of TO4 tetrahedra (T = Si , Al , and Be2+), which are linked at the corners and give rise to large cages (Fig. 1). These tetrahedra are fully ordered. The aluminosilicate framework of cubic sodalite, ideally Na8[Al6Si6O24]Cl2, is characterized by parallel six- membered rings consisting of alternating AlO4 and SiO4 tetrahedra. The cubic symmetry is the result of the stack- ing of such six-membered rings in an ABCABC... se- quence, which leads to large cages that are bounded by eight six-membered rings parallel to {111} planes and six four-membered rings parallel to {001} planes. The four-membered rings also consist of alternating AlO4 and SiO4 tetrahedra (Fig. 1a). The sodalite or ␤-cages 3+ enclose [Na4¥Cl] clusters (Figs. 1b, c). The structure of tugtupite (tetragonal), ideally Na8[Al2Be2Si8O24]Cl2, is isotypic with that of sodalite. The ␤-cages in tugtupite 3+ also enclose [Na4¥Cl] clusters, as in sodalite (Hassan & Grundy 1991). The sodalite-group minerals were studied by several researchers (e.g., Pauling 1930, Löns & Schulz 1967, Taylor 1967, 1968, Dan¿ 1966, Peterson 1983, Hassan & Grundy 1984, 1991). The helvite-group minerals, (Mn,Fe,Zn)8[Be6Si6 O24]S2, are also isotypic with sodalite. The Mn-rich member is helvite, the Fe-rich member is danalite, and the Zn-rich member is genthelvite. These minerals were also studied by several researchers (e.g., Hassan & Grundy 1985, Barth 1926, Burt 1980, Dunn 1976). The ␤-cages in the helvite-group minerals enclose clusters 6+ of [(Mn,Fe,Zn)4¥S] . Because these clusters are highly charged (i.e., 6+ valence units), they have smaller sizes 3+ than the [Na4¥Cl] clusters of sodalite and tugtupite (Hassan & Grundy 1985, 1991). The dimensions of the framework tetrahedra in the helvite-group minerals are constant regardless of the nature of the interframework

FIG. 1. (a) Projection of the unit cell of the sodalite structure cation because of the rotational freedom of the frame- down [001], drawn using the program STRUPLO. (b) The work tetrahedra (Hassan & Grundy 1985). In particu- 6+ truncated cubo-octahedral or ␤-cage in sodalite. (c) lar, pure danalite consists of 100% [Fe4¥S] clusters, 6+ Fourfold (trigonal pyramid) coordination of the Na site in and pure helvite consists of 100% [Mn4¥S] clusters. sodalite. These two minerals are investigated in this study.

163 40#1-fév-02-2328-11 164 3/28/02, 16:26 THERMAL ANALYSES OF SODALITE, TUGTUPITE, DANALITE AND HELVITE 165

EXPERIMENTAL position is given in Table 1. For the thermal analyses, 98.7 mg of finely powdered sodalite sample was used. The samples of sodalite, tugtupite, danalite, and The TG and DTA curves and their corresponding de- helvite were crushed, and pure crystals were handpicked rivative curves (DTG and DDTA, respectively) for so- and crushed to a powder using an agate mortar and dalite are shown in Figure 2. The temperature range pestle. A portion of the powder was used for the DTA from 20 to 850°C did not lead to any significant change and TG analyses, and another portion was used for high- in weight or peaks, and is therefore left out of Figure 2. temperature X-ray-diffraction (XRD) analyses. The re- The DTG and DDTA curves are obtained from the raw sults will be published elsewhere (Antao & Hassan, in TG and DTA data, respectively, using a narrow win- prep.). The samples used in this study were also used by dow for filtering the measured raw data. The differen- Hassan & Grundy (1984, 1985, 1991) to refine their tiation was done by using a modified GolayÐSavitzky crystal structures; they thus are well characterized. The algorithm of second order. Sodalite data obtained from composition of the samples used is presented in Table 1. these curves are summarized in Table 2. The experiments were carried out with a Netzsch Four peaks are observed in the DTA curve; peaks 1, STA 409 EP/3/D simultaneous TGÐDTA instrument. 3, and 4 are well defined in both the DTA and DDTA The data were analyzed using software programs sup- curves, but peak 2, although visually detectable, is less plied with the instrument (Hassan 1996). The thermo- obvious in the DTA curve, but is clearly seen in the DTG gravimetric (TG) curve was corrected for buoyancy curve (Fig. 2, Table 2). At 1079°C, peak 1 occurs only effect, and the differential thermal analysis (DTA) curve in the DTA trace, but there is no loss in weight in the was corrected for baseline effect. Corrections for buoy- TG curve. Peak 1 thus is related to melting (Fig. 2a). At ancy and baseline effects were obtained in a blank run 1347°C, peak 4 also occurs only in the DTA trace, and using empty crucibles that were later used to run the there is no loss in weight; we have no explanation for sample in a second run, but the two experimental runs the cause of peak 4 (Fig. 2a). After the experiment, the were made under identical conditions. The relationship residue in the crucible was found to be a porous powder between change in enthalpy and peak area in the DTA that was easily removed just by tapping the crucible on curve was determined by calibration using various stan- a desk. This is in contrast to the residue observe for the dard materials. other minerals, where the product of melting is a glass. Wellman (1970) studied the stability of sodalite in the RESULTS AND DISCUSSION range of 500Ð700°C and 600Ð2000 bars fluid pressure and indicated that sodalite is stable on the vapor-satu- Sodalite, Na8[Al6Si6O24]Cl2 rated liquidus, so sodalite can be a magmatic mineral. There are two well-defined DTG peaks, 2 and 3 The blue sodalite sample used in this study is from (Table 2, Fig. 2b). A continuous loss in weight occurs the Princess mine, Bancroft, Ontario. Its chemical com- over peaks 2 and 3 and corresponds to a net loss in

163 40#1-fév-02-2328-11 165 3/28/02, 16:26 166 THE CANADIAN MINERALOGIST

weight of about 11.5%. This continuous weight-loss can Halite, NaCl be somewhat arbitrarily divided into two steps, with a slow loss of about 4.5 wt.% for peak 2, and a more rapid For the thermal analyses of halite, 43.0 mg of the loss of about 7.0 wt.% for peak 3 (Fig. 2b). In the so- sample was used. The thermal curves for halite are dalite formula, Na6[Al6Si6O24]¥2NaCl, the 2NaCl com- shown in Figure 3, and the data from these curves are ponent constitutes 12.1 wt.%, which is comparable to summarized in Table 3. Three peaks are observed in the the total loss in weight of about 12.2% obtained experi- DTA curve (Figs. 3a, c); peaks 1 and 3 are well defined mentally (Table 2). According to the chemical data on in both the DTA and DDTA curves, but peak 2, although sodalite, Cl constitutes 7.17 wt.% (Table 1), so 4.93 visually detectable, is less obvious in the DTA curve, wt.% loss of Na is required for the total weight-loss to but is clearly seen in the DTG curve. At 803°C, peak 1 be about 12.1%. Moreover, the TG results indicate that occurs only in the DTA trace, and there is no loss in NaCl was lost at two different rates. To investigate the weight. Peak 1 is thus attributed to the melting of NaCl. results of sodalite further, halite was thermally analyzed. This melting point is very close to that reported, 800°C, in the literature (e.g., Neuman 1977). Peaks 2 and 3 in the thermal curves correspond to loss of volatiles. The TG curve of NaCl shows that prac- tically all the mass is lost (Table 3). A continuous loss of weight occurs over peaks 2 and 3 and corresponds to a net loss in weight of about 98.5%. This continuous weight-loss may somewhat arbitrarily be divided into two steps, with losses of about 38.7 wt.% for peak 2 and 59.5 wt.% for peak 3 (Fig. 3b); the loss of NaCl is slower in the first step compared to the second step. The DTA and TG results for halite and sodalite are similar to each other with regard to the three peaks (1, 2, and 3). In both samples, NaCl is liberated at two different rates. Peaks 2 and 3 in both samples are attributed to the loss of NaCl at different rates; peak 2 corresponds to a slow rate of loss, and peak 3, to a more rapid rate of loss.

FIG. 2. Thermal curves for sodalite: (a) TG and DTA curves, (b) TG and DTG curves, and (c) DTA and DDTA curves. Corresponding peaks at a particular temperature are given the same number and are labeled on the DTA and DTG curves in this and the other figures that contain thermal curves.

163 40#1-fév-02-2328-11 166 3/28/02, 16:26 THERMAL ANALYSES OF SODALITE, TUGTUPITE, DANALITE AND HELVITE 167

Tugtupite, Na8[Al2Be2Si8O24]Cl2 peak 2 in the DTA curve also encloses peak 1, which corresponds to a weight-loss step. Thus, peaks 1 and 2, The red tugtupite sample is from Ilímaussaq, Narssaq which are well separated in sodalite, overlap in tugtupite Kommune, South Greenland (Royal Ontario Museum, (Figs. 2a, 4a). It should be noted that both the DTA and ROM #M32790). Table 1 contains the chemical data for DTG curves are mirror images of each other. However, a sample from the same locality (Dan¿ 1966). For the the DTG curve only contains peaks corresponding to thermal analyses, 103.0 mg of the finely powdered weight loss, whereas the DTA curve contains these tugtupite sample was used. The thermal curves for peaks in addition to peaks corresponding to transitions. tugtupite are shown in Figure 4. The temperature range Signs of melting (i.e., glass) were observed in the from 20 to 850°C did not show any significant change crucible after the experiment. in weight or peaks, and are therefore left out of Figure 4. Several DTG peaks were observed (Fig. 4b). The Data for tugtupite obtained from these curves are sum- weight loss can be divided into two main steps, with marized in Table 4. Many peaks are observed in the losses of about 1.8 wt.% for peak 1 and 8.2 wt.% for all DTA and DDTA curves. At 1029°C, peak 2 occurs only the other peaks beyond 1250°C. All the TG peaks cor- in the DTA trace, and there is no loss in weight in the respond to a net loss in weight of about 10.0%. These TG curve. Peak 2 thus is related to melting. Note that weight losses are attributed to the escape of NaCl. Ac- cording to the chemical analyses, Cl constitutes 7.28 wt.% (Table 1), so 2.72 wt.% loss of Na is required for the total weight loss to be about 10.0%. In the tugtupite formula, Na6[Al2Be2Si8O24]¥2NaCl, theoretically the 2NaCl component constitutes 12.49 wt.%, which is comparable to the net loss in weight of about 10.0% obtained from the TG curve of tugtupite. Note that at the end of the experiment, weight loss was still taking place, as indicated by peak 4 in the DTG curve (Fig. 4b). However, this weight loss is not reflected in the corrected TG curve because the blank correction curve was truncated before 1450°C, but the uncorrected TG curve (not shown) indicates the corresponding weight-loss peaks 4a, b; these data are included in Table 4. In tugtupite, each weight-loss step is attributed to NaCl, escaping at different rates. The tugtupite sample was run a second time, and the results obtained were the same. In both tugtupite and sodalite samples, NaCl is liberated at different rates. In sodalite, the loss of NaCl

FIG. 3. Thermal curves for halite: (a) TG and DTA curves, (b) TG and DTG curves, and (c) DTA and DDTA curves.

163 40#1-fév-02-2328-11 167 3/28/02, 16:26 168 THE CANADIAN MINERALOGIST

occurs in two continuous steps; in the first step, the loss atoms (Al, Be, Si) compared to two different types of T of NaCl is slow, but in the second step, it is more rapid. atoms (Al, Si) in sodalite. The temperature at which However, in tugtupite, NaCl escapes in several steps at tugtupite (926°C) begins to lose NaCl is about the same different rates. This difference in behavior may be re- as that in sodalite (933°C), but the loss from the sodalite lated to the number of bonds with different strengths melt is stopped at about 1384°C, whereas in the that have to be broken to liberate NaCl. In sodalite, there experiment on the tugtupite, the melt continues to lose are three NaÐO bonds of equal strength that have to be NaCl beyond 1442°C. broken, but in tugtupite there are four NaÐO bonds with different strengths that have to be broken to liberate NaCl (Fig. 5). Moreover, the escape of NaCl from sodalite may occur more easily because of the larger ␤- cages (V = 701 Å3) in sodalite compared to those in tugtupite (V = 662 Å3). In tugtupite, melting occurs at 1029°C, which is 50°C lower than in sodalite. This difference in temperature may be attributed to more strain in the tugtupite framework, which contains three different types of T

FIG. 4. Thermal curves for tugtupite: (a) TG and DTA curves, (b) TG and DTG curves, and (c) DTA and DDTA curves.

163 40#1-fév-02-2328-11 168 3/28/02, 16:26 THERMAL ANALYSES OF SODALITE, TUGTUPITE, DANALITE AND HELVITE 169

Danalite, Fe8[Be6Si6O24]S2 Peaks 1, 2, and 3 in the DTA trace correspond to various rates of increase in weight as indicated by the A sample of danalite from Mt. Francisco granitic TG curve (Fig. 6). Peaks 4, 5, 6, and 7 in the DTA curve pegmatite, Ribawa area, Western Australia (ROM are attributed to the loss of volatiles at different rates #M37261) was used is this study. The chemical data for because of the continuous decrease in weight in the TG a sample from the same locality are given in Table 1. curve. The onset and end temperatures for all the peaks This material was studied by Hassan & Grundy (1985). are better defined by the DDTA curve. Peak 7 at about For the thermal analyses, 102.0 mg of the finely pow- 1274°C in the DTA trace is also related to the melting dered sample was used. The thermal curves for danalite of danalite and is close to the end of the weight-loss are shown (Fig. 6). In general, the TG curve indicates event. Melt was observed in the crucible after the ex- three main events, first a weight-gain event, followed periment. by a weight-loss event, and finally another small weight- Peaks 1, 2, and 3 in the DTG curve correspond to gain event (Fig. 6a). The other thermal curves (Fig. 6) the first weight-gain stage, and from the TG curve a net are more complex and contain many peaks, unlike the gain of 4.0 wt.% is obtained (Fig. 6, Table 5). The gain TG curve. Data obtained from these curves are summa- in weight results from oxidation because the sample was rized in Table 5. heated in a static air environment. Therefore, divalent Fe2+ and Mn2+ oxidize to trivalent Fe3+ and Mn3+, respectively, at different rates. With complete oxidation, the theoretical formula of danalite (initially consisting 2+ 2+ only of Fe ) would change from Fe 8[Be6Si6O24]S2 to 3+ Fe 8[Be6Si6O28]S2, and the theoretical increase due to the extra four O atoms would be 5.7 wt.%. If the theore- 2+ 3+ tical formula changed to Fe 2Fe 6[Be6 Si6O27]S2, the theoretical increase due to the extra three O atoms would be 4.3 wt.%. From the TG curve, a weight gain of 4.0 wt.% is obtained, which corresponds to a theoretical increase of 2.8 O atoms. This sample of danalite contains about equal amounts of Fe and Mn, and a small

FIG. 5. Na atom coordinations: (a) five coordination in FIG. 6. Thermal curves for danalite: (a) TG and DTA curves, tugtupite, and (b) four coordination in sodalite. (b) TG and DTG curves, and (c) DTA and DDTA curves.

163 40#1-fév-02-2328-11 169 3/28/02, 16:26 170 THE CANADIAN MINERALOGIST

amount of Zn, which will not be oxidized (Table 1). The Fe2+ first begins to oxidize to Fe3+ because of its lower oxidation potential (+0.77 V for Fe2+ → Fe3+ and +1.51 V for Mn2+ → Mn3+; Bard et al. 1985). Weight loss occurs over peaks 4, 5, 6, and 7 in the DTG curve, and from the TG curve a net loss of 5.7 wt.% is obtained (Fig. 6, Table 5). Sulfur constitutes 5.73 wt.% of danalite (Table 1), which is comparable to the weight loss obtained from the TG curve. Therefore, this weight loss is attributed to the loss of S2(g) at different rates. In the weight-gain stage, O atoms are incorporated into the danalite structure, and in the succeeding weight-loss stage, instead of losing the incorporated O atoms, S atoms are lost as S2(g). Beyond about 1297°C, the TG curve begins to increases in weight again (incomplete peak 8 in the TG curve, Fig. 6a). An increase of 0.7 wt.% is obtained up to 1450°C (Table 5). Possibly the trivalent Mn3+ begins to oxidize to tetravalent Mn4+. The peak temperature for this weight gain is beyond the experimental limit of 1450°C; therefore, the DTG or DTA curve did not give a peak to indicate this increase in the TG curve. In danalite, there is a weight gain caused by oxidation of (Fe, Mn)2+ to (Fe, Mn)3+ followed by a loss of S2 (g), melting, and finally another weight-gain stage, possibly caused by further oxidation of Mn3+ to Mn4+ in the melt. During the first weight-gain event, which begins at about 771°C, the O atoms incorporated possibly enter the cubo-octahedral cavities, and occupy a general site such that they are within bonding distance to the Mn or Fe atoms. Such a general site was found for the O atoms of OH and H2O in basic sodalite (Hassan & Grundy 1983). If the Mn and Fe atoms fully oxidize to (Fe, Mn)3+, two O atoms are expected to enter each cubo-octahedral cavity. Thus, the original four-fold coordination of the Mn or Fe atom (tetrahedrally coordi- nated to three O atoms of the framework and one central S atom), probably increases to six, and the structure of danalite still remains intact. From about 1029°C, each Mn or Fe atom, which is now bonded to one central S atom, three O atoms from the framework, and two in- 2+ 6+ 3+ corporated O atoms, cannot retain the coordination num- hedra change from [(Mn,Fe) 4¥S] to [(Mn,Fe) 4 6+ ber of six. The distance of the interframework cation, ¥SO2] in the first stage of oxidation, with the net IC (IC = Mn, Fe) to S, is longer than the ICÐO bond charge on the cage cluster remaining at +6 valence units. distance [IC–S = 2.413(1) Å, IC–O = 2.033(3) Å; When danalite loses all of the S atoms, the cage clusters 3+ 8+ Hassan & Grundy 1985]. Therefore, the longer ICÐS change to [(Mn,Fe) 4¥O2] . In this case, the net charge bonds are broken, and S2 (g) is liberated. The coordina- on the clusters (+8 valence units) does not balance the tion of the IC cation is now five, all the anions being net change on the framework (Ð6 valence units), so the oxygen atoms. The incorporated O atoms, however, do danalite structure becomes unstable and the sample not occupy the positions of the S atoms, i.e., at the cen- melts. To balance the excess of +2 valence units, more ter and corners of the unit cell, because the ICÐS bond O atoms are required. Therefore, with further oxidation in danalite is too long [IC–S = 2.413(1) Å] to allow for of the melt (Mn3+ → Mn4+), the O atom content in- reasonable ICÐO bond distances. The incorporated O creases, the charge balances, and the melt becomes stable. atoms are probably located off the center of the unit cell and are closer to the IC cation to give proper ICÐO bond Helvite, Mn8[Be6Si6O24]S2 distances. More O atoms can be incorporated into the danalite-derived melt with further oxidation of Mn3+ to The helvite sample is from Sunnyside, San Juan Mn4+. Thus, the contents of some of the cages of tetra- County, Colorado (ROM #M36390). In the thermal ex-

163 40#1-fév-02-2328-11 170 3/28/02, 16:26 THERMAL ANALYSES OF SODALITE, TUGTUPITE, DANALITE AND HELVITE 171

Therefore, similar to the case for danalite, helvite un- dergoes oxidation, and contains extra incorporated O FIG. 7. Thermal curves for helvite: (a) TG and DTA curves, atoms. The TG curve gives a total weight loss of 4.7 (b) TG and DTG curves, and (c) DTA and DDTA curves. wt.% over peak 2. This weight loss is attributed to the loss of S2 (g), as in danalite. S constitutes 5.78 wt.% in the theoretical chemical composition of helvite (Table 6), but from the TG curve a weight loss of 4.7 wt.% is periment, only 20.7 mg was used instead of about 100 obtained. Beyond about 1333°C, the TG curve shows a mg of helvite, as more sample was unavailable. The small gain in weight of 0.2 wt.%. At the maximum tem- theoretical chemical composition of pure helvite, perature, the DTG and DDTA curves also seem to indi- Mn8[Be6Si6O24]S2, is given in Table 1. The thermal cate peaks corresponding to this weight-gain stage. curves for helvite are shown in Figure 7. Data obtained However, since the experiment was terminated at from these curves are summarized in Table 6. In gen- 1450°C, this weight-gain stage is not well defined. eral, from the TG curve, there are two main stages; a Therefore, in this minor weight-gain stage, possibly weight-gain stage is followed by a weight-loss stage, as Mn3+ further oxidizes to Mn4+. was found also in danalite. Two sharp peaks (1 and 2) The DTG curve contains small peaks in addition to in the DTG curve correspond to these two events. The the two major peaks 1 and 2 (Fig. 7b); however, these DTA curve contains three small peaks (1, 2 and 3) pos- small undulations are not really peaks, but noise, which sibly because of the small amount of sample. Peaks 1 arises because of the small amount of sample used (20.7 and 2 in the DTA curve are related to weight gain and mg). Larger filters (e.g., 101p) can smooth the DTG weight loss, respectively, because of the change in curve and eradicate most of the noise, but the peak tem- weight in the TG curve. Peak 3 occurs in the DTA trace peratures obtained would not be accurate, so the use of at about 1260°C, and there is no loss in weight in the larger filters was avoided. TG curve; peak 3 thus is related to melting. In both danalite and helvite, first there is a weight- Peaks 1 and 2 are well defined in the DTG curve. gain stage, which begins at a temperature of about The TG curve shows that there is a weight gain of 5.0 705°C in helvite and about 771°C in danalite. This wt.% over peak 1 (Fig. 7, Table 6). This gain in weight weight gain is caused by the oxidation of (Fe, Mn)2+ to 2+ 2+ 3+ is caused by the oxidation of divalent Mn and Fe to (Fe, Mn) . Following this event there is a loss of S2 trivalent Mn3+ and Fe3+, respectively. With complete (g), which begins at 883°C in helvite and 1029°C in oxidation, the theoretical formula of pure helvite (con- danalite. The weight gain and weight loss in danalite 2+ sisting only of Mn) would change from Mn 8[Be6Si6 occur in several steps and at different rates, unlike those 3+ O24]S2 to Mn 8[Be6Si6O28]S2; the theoretical increase observed in helvite. The weight-gain stage in danalite due to the extra four O atoms is 5.8 wt.%. From the TG was expected to occur at a lower temperature than that curve, a weight gain of 5.0 wt.% is obtained. This for helvite because of the lower oxidation potential of indicates a theoretical increase of 3.4 extra O atoms. Fe2+ compared to that of Mn2+. However, the onset tem-

163 40#1-fév-02-2328-11 171 3/28/02, 16:26 172 THE CANADIAN MINERALOGIST

peratures for both the weight loss and weight gain in HASSAN, I. (1996): The thermal behavior of cancrinite. Can. helvite are lower than those observed for danalite. Melt- Mineral. 34, 893-900. ing occurs at about 1260°C in both samples. Finally, beyond about 1300°C there is further weight-gain, pos- ______& GRUNDY, H.D. (1983): Structure of basic sodalite, sibly caused by further oxidation of Mn3+ to Mn4+ in Na8Al6Si6O24(OH)2¥2H2O. Acta Crystallogr. C39, 3-5. both melts formed. As helvite is Mn-rich, a larger ______& ______(1984): The crystal structures of weight-gain is expected for helvite in the second stage sodalite-group minerals. Acta Crystallogr. B40, 6-13. of oxidation, compared to danalite, because only the Mn3+ (not Fe3+) can further oxidize to the +4 oxidation ______& ______(1985): The crystal structures of helvite state. However, in this second stage of oxidation, the group minerals, (Mn,Fe,Zn)8(Be6Si6O24)S2. Am. Mineral. observed weight-gain in danalite (0.7 wt.%) is more than 70, 186-192. that observed in helvite (0.2 wt.%). The mechanism of incorporating O atoms into the structure, and losing S ______& ______(1991): The crystal structure and atoms, is the same in both danalite and helvite. Finally, thermal expansion of tugtupite, Na8[Al2Be2Si8O24]Cl2. Can. Mineral. 29, 385-390. this study shows that thermal techniques may be used to determine volatiles, oxidation states, and transition LÖNS, J. & SCHULZ, H. (1967): Strukturverfeinerung von temperatures in minerals. This type of study may also Sodalit, Na8Al6Si6O24Cl2. Acta Crystallogr. 23, 434-436. have implications for studies concerning the degassing of magmas. NEUMAN, B.S. (1977): Thermal techniques. In Physical Methods in Determinative Mineralogy (J. Zussman, ed.). ACKNOWLEDGEMENTS Academic Press, London, U.K. (605-662).

We thank the reviewers, Prof. Igor Khodakovskii, PAULING, L. (1930): The structure of sodalite and helvite. Z. Kristallogr. 74, 213-225. Drs. Irina Kiseleva, Elena Sokolova, and Robert F. Martin for useful comments. PETERSON, R.C. (1983): The crystal structure of hackmanite, a variety of sodalite from Mont St-Hilaire. Can. Mineral. 21, REFERENCES 549-552.

BARD, A.J., PARSONS, R. & JORDAN, J. (1985): Standard TAYLOR, D. (1967): The sodalite group of minerals. Contrib. Potentials in Aqueous Solutions. IUPAC, Marcel Dekker, Mineral. Petrol. 16,172-188. New York, N.Y. ______(1968): The thermal expansion of sodalite group of BARTH, T.F.W. (1926): Die kristallographische Beziehung minerals. Mineral. Mag. 36, 761-769. zwischen Helvin and Sodalit. Norsk Geol. Tidsskr. 9, 40- 42. WELLMAN, T.R. (1970): The stability of sodalite in a synthetic syenite plus aqueous chloride fluid system. J. Petrol. 11, BURT, D.M. (1980): The stability of danalite, Fe4Be3(SiO4)3S. 49-71. Am. Mineral. 65, 355-360.

DAN¯, M. (1966): The crystal structure of tugtupite Ð a new mineral, Na8[Al2Be2Si8O24](Cl,S)2. Acta Crystallogr. 20, 812-816.

DUNN, P.J. (1976): Genthelvite and the helvine group. Mineral. Received February 11, 2001, revised manuscript accepted Mag. 40, 627-636. January 7, 2002.

163 40#1-fév-02-2328-11 172 3/28/02, 16:26